A supervised approach to windowing detection on dynamic networks

For any stream of time-stamped edges that form a dynamic network, a necessary and important choice is the aggregation granularity that an analyst uses to bin the data at. While this choice is often picked by hand, or left up to the technology that is collecting the data, the choice can make a big difference in the properties of the network. We call this the windowing detection problem. In previous work, this problem is often solved with a heuristic as an unsupervised task. As an unsupervised problem, it is difficult to measure how well a windowing algorithm performs. In addition, we show that the best windowing is dependent on which task an analyst want to perform on the network after windowing, and that therefore the task should be taken into account. We introduce a framework that tackles both of these issues: By measuring the performance of the windowing algorithm based on how well a given task is accomplished on the resulting network, we are for the first time able to directly compare different windowing algorithms to each other. Using this framework, we introduce windowing algorithms that take a supervised approach: they leverage ground truth on training data to find a good windowing of the test data. We compare the supervised approach to previous approaches and several baselines on real data. ACM Reference format: Benjamin Fish and Rajmonda S. Caceres. 2016. A supervised approach to windowing detection on dynamic networks. In Proceedings of ACM Conference, Washington, DC, USA, July 2017 (Conference’17), 10 pages. DOI: 10.1145/nnnnnnn.nnnnnnn